🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA DGX Spark run Megrez-3B-Omni?
Yes — here's how fast and up to what context, with real measured numbers.
Megrez-3B-Omni on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1643k | 33.2-39.6 t/s | ~1.2-1.5 (slow) |
| Q8 | yes | 1689k | 66.4-79.2 t/s | ~2.4-2.9 (slow) |
| Q4 | yes | 1709k | 118.6-141.4 t/s | ~4.3-5.2 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA DGX Spark run Megrez-3B-Omni?▶
Yes. At Q4 it generates ~118.6-141.4 tok/s and fits up to 1709k context; beyond that it spills to SSD swap and slows to ~4.3-5.2 tok/s.
How much context fits?▶
At Q4, up to about 1709k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.